TReC: Transferred ResNet and CBAM for Detecting Brain Diseases

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Neurology

Background:

• Early diagnosis of brain diseases is crucial for timely intervention and improved patient outcomes.
• Classifying brain diseases using medical images presents challenges due to data scarcity and computational demands.

Purpose of the Study:

• To develop an efficient brain disease detection model for small-scale magnetic resonance (MR) datasets.
• To introduce a novel approach, TReC (Transferred Residual Networks-Convolutional Block Attention Module), for enhanced brain MRI classification.

Main Methods:

• Utilized a pre-trained ResNet model initialized with ImageNet weights.
• Integrated the Convolutional Block Attention Module (CBAM) into ResNet residual blocks.
• Replaced fully connected layers and retrained the entire TReC model on brain MR datasets.

Main Results:

• The TReC model demonstrated superior performance in both two-class and multi-class brain disease classification tasks.
• Achieved state-of-the-art results compared to existing models on evaluated MR datasets.

Conclusions:

• The proposed TReC model is effective for brain disease detection, particularly in scenarios with limited sample sizes.
• This approach offers a promising solution for improving the accuracy and efficiency of brain disease classification from MR images.